JP4673411B2 - Method and apparatus in a mobile communication network - Google Patents

Description

  The present invention relates generally to packet-based communication systems suitable for acoustic signal transmission, and more particularly to buffer techniques for use in such communication systems.

  Voice over IP is an aggregation between the world of communication and data communication, and a call signal is carried by a data packet, for example, an Internet Protocol (IP) packet. The call to be recorded is encoded on a frame-by-frame basis using a call codec. A data frame is generated for each call frame. One or several data frames are grouped into RTP packets. RTP packets are further combined into UDP packets, and UDP packets are combined into IP packets. The IP packet is then transmitted from the sending client to the receiving client using the IP network.

  A problem associated with packet-based networks is delay jitter. Delay jitter means that even if packets are sent at regular intervals, for example, one frame every 20 milliseconds, the packets arrive irregularly at the receiver. Packets may arrive out of order. The most common reason for receiving packets out of order is because, at least in the case of fixed networks, packets take different routes. For wireless networks, another reason may be to use retransmission. When transmitting packet N on the uplink (ie, from the mobile terminal to the base station), there may be bit errors that cannot be corrected, requiring retransmission to be performed. On the other hand, the signal transmission for retransmission is slow, so the next packet (packet N + 1) waiting for the turn is transmitted, and then packet N is retransmitted. This would result in receiving packets out of order if packet N + 1 was correctly received before correctly receiving retransmission packet N.

  In a VoIP client, jitter buffer means are used to equalize transmission delay jitter, so that speech samples can be played out at a constant sample rate, eg, one frame every 20 ms (the playout is described herein). Is used to indicate the transmission of a call to the sound card). The full level of the jitter buffer means is proportional to the amount of delay jitter in the packet flow, and the objective is to keep the amount of delay loss at an acceptable level while keeping the delay as small as possible. The following example illustrates the importance of keeping the delay as low as possible: A long buffer time in the jitter buffer means increases the end-to-end delay. This may lead to a decrease in conversation quality because the system feels "slow". A large delay increases the risk of the user speaking at the same time, and may give the impression that the other user is “slow” (think slowly). In addition, a packet with a delay loss is received correctly, but arrives too late to be useful to the decoder.

  The jitter buffer means accumulates packets or frames for a fixed time. A typical way to define this is to fill the jitter buffer means to a certain “level”, eg shown as a full level. The reason that this level is often measured in milliseconds instead of the number of frames is because the frame size varies. Therefore, the level of the jitter buffer means is measured in time. The level of the jitter buffer means can be set in several different ways.

Fixed size: Fixed size means that the jitter buffer fullness level is fixed and pre-configured. Before restarting call playout after the DTX period, the jitter buffer means is first filled with a fixed time, eg, a fixed number of frames (eg, 5 frames). This initial margin is used to provide protection against delay jitter and delay loss.
Adaptive size of jitter buffer means: The full level of the jitter buffer varies with delay jitter. As in the case of the full size of the fixed size jitter buffer, the initial frame number is buffered, and then the call playout is resumed after the DTX period. Nevertheless, during voice operations (non-DTX), the full level of the jitter buffer means may change based on the analysis of the incoming packet. Statistics can be aggregated through several conversations. On the other hand, at the start of each call, the full level of the jitter buffer is normally reset to “unspecified level”.

  Adaptive size of jitter buffer means to improve interoperability: In order to reduce the perceived delay, it is possible to initialize the jitter buffer means in a shorter time than with the adaptive size of the jitter buffer means, after DTX As soon as the first call packet is received, call playout is started. In order to reach the full level of the jitter buffer, a time measurement is used to stretch the first decoded frame and reduce the pace to extract packets from the jitter buffer means. Time measurement means that the call frame is adaptively played out, i.e., the call frame that normally contains a 20 ms call can be expanded, producing a 30 ms call. An alternative to starting playout after the first received packet is to wait for one or two extra packets. Patent Document 1 and Patent Document 2 describe time measurement.

  DTX is discontinuous transmission, which means that when there is no voice and the input signal contains only (background) noise, a special type of information is sent on the channel. The encoder evaluates the background noise and determines a set of parameters that describe the noise (= silence description (SID) parameters). The SID parameter is transmitted to the receiving terminal, and similar noise and comfort noise can be generated. SID parameters are transmitted less frequently than normal call frames, saving power and transmission resources.

  Referring now to FIG. 1, FIG. 1 shows an example of the operation of the initial jitter buffer means by the adaptive size method of the jitter buffer means for improving the interoperability. The top diagram shows the jitter buffer fullness level and the bottom diagram shows the frame size. As soon as the first packet is received, playout starts in about 0.5 seconds. Time measurement is performed to increase the size of the generated frame and process the frame from the jitter buffer means slower than the normal pace. The early start of playout gives a feeling of improved interoperability and an improvement in conversation quality. At the end of the conversation burst, the last call frame is shortened and played out at a faster pace than usual at about 3 seconds. This further improves interoperability.

  Although the adaptation of the target jitter buffer means level (60 ms) during non-DTX periods is not shown in FIG. 1, a typical implementation for the adaptive size of the jitter buffer means to improve interoperability has this capability. Note that it will exist.

  On the other hand, the above three methods have several drawbacks. Since several packets are always buffered before the playout starts, the size of the fixed jitter buffer means introduces a considerable delay. This makes the user feel a decrease in interoperability.

  If at least the channel fluctuates slowly, the adaptive jitter buffer means can adjust the full level to introduce less delay on average. The problem of poor interoperability due to the long initial buffer time still remains, because the adaptation purpose is to adapt in the ongoing packet flow during call activity if the flow starts after the DTX period. It should be noted that this problem occurs if the jitter buffer full level is reset to an unspecified level at the beginning of each call (ie, when switching from DTX to call).

  Initialization of the jitter buffer means improves interoperability, since an initial delay will be less felt when using an adaptive size of the jitter buffer means that improves interoperability. However, one problem is that the level of the jitter buffer means is very low at the start of a call burst, so there is a risk that the delay jitter at the start of the call burst will become a delay loss. Similar to frame loss, delay loss will reduce call quality because it is lost or triggers error concealment of later received frames. In addition, the adaptive sizing method of the jitter buffer means to improve interoperability adjusts the buffer level to the normal full level, as the adaptation period must be shortened sufficiently to avoid encountering multiple delay spikes It also means that time measurement must be done fairly quickly. A delay spike is when the delay increases considerably from the first packet to the subsequent packet. This means that timekeeping must be fairly aggressive. Aggressive timing increases the risk that the timing itself introduces distortion. Distortion can include various types of clicks, props, and noise bursts, but can also include “strange sounding sounds” such as “unnatural conversation volume”.

  Most modern speech codecs (GSM-EFR, GSM-AMR, ITU-TG.729, EVRC, etc.) that use inter-frame prediction that allows coding of signals that maintain low bit rate but maintain quality In the case of, there is a further problem. Since the error propagates for some time due to inter-frame prediction, both frame loss and delay loss also distort the current frame and also subsequent frames. The error propagation time depends on the sound and codec, but can be as long as 5-6 frames (100-120 ms). Delay loss is particularly significant at the beginning of a speech burst because these parts often include the start of utterance and the adaptive codebook later uses the start of utterance to construct the speech waveform. The consequences of delay loss at the start of a call burst are therefore often very audible and can significantly impair understanding.

  Although there are several ways to compensate for error propagation that can occur if delay loss occurs during the rise time, they all have significant drawbacks. One possibility is to reduce the initial buffer time, but not much at best. This will of course mean that it is not possible to obtain as much gain as desired for interoperability.

  Another possibility is to reduce the interframe prediction used for the codec. Nonetheless, this will either reduce the original call quality because inter-frame correlation is not used to its full potential, or it will be necessary to encode the signal at a higher bit rate, Or both.

Due to the disadvantages of the adaptive sizing method of the jitter buffer means that improves interoperability, it is difficult to use this method in real systems. This method may work well for channels that contain very little jitter, and preferably also with little packet loss, but for channels that contain lots of jitter and possibly also cause packet loss, it can improve interoperability. It is very difficult to obtain a sufficient gain. In most practical cases it may be preferable to have several frame initialization times before the playout begins.
International Publication No. 2001118790 A1 Pamphlet US Patent Application Publication No. 2004 / 0156397A1

  It is an object of the present invention to achieve control logic means that improve interoperability and / or call (listening) quality.

  The above objective is accomplished by control logic means and methods according to the independent claims.

  Preferred embodiments are defined by the dependent claims.

  The present invention improves interoperability and / or call and listening quality based on the possibility of adaptively controlling at least one of initial buffer time and time measurement.

  This is achieved by the introduction of control logic means for extracting information from at least one of the jitter buffer means, the decoder, the time measuring means, and the state return means, the control logic means being based on the initial buffer time and the time based on the information to be extracted. At least one of the measured quantities is further adaptively controlled.

  By introducing the control logic means, it is possible to improve the advantages of the state return means in combination with the initial buffer time. State recovery reduces the receiver's sensitivity to delay loss during the initial rising edge of the jitter buffer means. Therefore, by having an active time measurement, it is possible to have a very short initialization time. This further improves interoperability over what is possible with the adaptive size method of jitter buffer means to improve interoperability.

  Since the strength against delay loss is increased by the state recovery means, a longer rise period of the jitter buffer means can also be tolerated. Therefore, it is possible to measure the time to reduce the aggressiveness. This can be advantageous because time measurements can introduce different distortions into different calls in different sounds depending on the time measurement performance.

  Since the control logic means can combine the initial buffer time, state recovery and time measurement in various ways, adaptation between these variables improves performance. This adaptation can be based on either current channel conditions or acoustic signals, or both.

  The use of time measurement and state recovery adds complexity and therefore makes the central processing unit (CPU) more burdensome. A further advantage with the present invention is that the present invention allows control of complexity by controlling the enabling / disabling of parameter setting and state return. This is achieved by control logic means that retrieves information about the CPU load and controls parameter setting or enabling / disabling of the state return means. Information taken out regarding the CPU load can be associated with at least one of the time measurement operation and the state return operation.

  Reduces the delay felt to improve interoperability, thus improving listening quality and repairing possible delay losses through active use of time measurement, thus improving quality by returning to the state.

  Another advantage is that the control logic means good performance in heavily loaded cell networks with large jitter and possibly also packet loss and large delays, sometimes in office LANs with delay spikes and short delays or bad channel conditions. It is possible to adapt to various operating conditions such as channel conditions.

  Objects and advantages of the present invention will become apparent from a reading of this specification in conjunction with the drawings.

  The invention will now be described more fully with reference to the accompanying drawings, which illustrate preferred embodiments of the invention. On the other hand, the invention may be implemented in many different forms and should not be construed as limiting the invention to the embodiments shown herein; rather, the present disclosure is complete and complete. As presented, the present invention will fully convey the scope of the invention to those skilled in the art.

  An overview of a receiver for a voice client over IP is shown in FIG. It should be noted that the present invention may also be applicable to voice over ATM without IP and other systems where delay jitter occurs. The receiver includes jitter buffer means connected to the decoder, the decoder further connected to time measuring means. The state recovery means connected to the error concealment (ECU) means can be connected to the jitter buffer means and the decoder. The receiver receives the data packet via the jitter buffer means. Since a packet can contain data for several call data frames, if the packet is depacketized and contains data for several call data frames, this is indicated in the packet pay header. Therefore, the jitter buffer means extracts a data frame from the packet. Since the data frames can be reordered due to delay jitter, the jitter buffer means keeps the frames in order. From the jitter buffer means it is possible to obtain information about delay loss, frame loss and the current level of jitter buffer means. It should be noted that the extraction of the data frame can be performed in another means, in which case the jitter buffer means receives the data frame.

  The decoder decodes the data frame into a speech frame, i.e. an acoustic signal. For example, the data frame is decoded into 160 16-bit word samples (speech frames) as in the AMR 12.2 kbps mode where the data frame is 244 bits.

  The time measuring means has the ability to compress or decompress the size of the decoded speech frame input from the decoder, for example 160 samples from the speech decoder can be decompressed to 240 samples or compressed to 80 samples It can be expanded / compressed to other frame sizes. The time measurement means provides information regarding the compression or expansion to be achieved. Time measurement is performed differently for various acoustic signals. Some acoustic signals are very easy to measure time, in which case the time measurement introduces little or no distortion. Examples of such sounds are unchanged voiced segments, unvoiced segments and background noise. Other acoustic signals are extremely difficult to measure time, and in this case, the time measurement will probably introduce distortion that is virtually audible. Examples of acoustics that are difficult to measure are instantaneous (transition from unvoiced to voiced), plosive ("t", "p", "b", ...), call start (call from background noise (DTX)) ). Therefore, in order to be able to determine the aggression that should be timed, it is desirable to retrieve information from either the call decoder or the timed function. Other examples of various parameters or measurement scales that may be taken alone or in combination to describe acoustic or channel characteristics and to adapt to initial buffer time or time measurement are adaptive codebooks. , ACB) gain, fixed codebook (FCB) gain, LTB / ACB delay measurement measure and characteristic, LSP rate characteristic, spectral flatness measurement measure, spectral deviation, energy measurement measure and deviation.

  Various example methods of performing time measurements that can be used in the present invention are described in patent applications, WO 93516 and US Pat. No. 6,873,954. It should be noted that the above time measuring means can also be located at the transmitter, so that the time measurement can be performed before the encoding operation. If time measurement is performed at the transmitter, some information must be exchanged between the encoder and decoder.

  The decoding means and the time measuring means can also be integrated into one device. In such an integrated device, time measurement is activated and performed in front of the synthesis filter, and then the synthesis filter is applied to more or fewer samples than usual.

  A time measuring means is further connected to the sample buffer. The time-measured frame is transmitted to the sample buffer. If the sample buffer is full to a pre-defined threshold level, one or more samples, the frame being a plurality of consecutive samples, are transmitted from the sample buffer to the speaker's acoustic card. If the sample buffer is not full, further decoding operations are required. There can therefore be the introduction of time measurement.

According to the present invention, control logic means is introduced to extract information such as information about the existing state recovery function, channel characteristics, acoustic characteristics, induced distortion (comparison of distortion before and after time measurement) and execution time measurement. Information about the existing state return function can be preconfigured in the control logic means, i.e. information about enabling / disabling of the state return means can be derived from the state return means. Channel characteristics can be derived from the jitter buffer means, acoustic characteristics and distortion information can be derived from the decoder, and distortion information and implementation time measurements can be derived from the time measurement means. Therefore, in order to have information about whether or not the state return function is available, control logic means is required, and the control logic means receives information from at least one of the jitter buffer means, the state return means, the decoder, and the time measurement means. Try to take it out. Control logic means can also be used to control the full level of the jitter buffer.

Then at least one of the initial buffer time of the jitter buffer means based on the information retrieved from the jitter buffer means and the information on the availability of the state recovery function and the time measurement setting of the time measurement means based on the information retrieved from the time measurement means or the decoder, Control logic means adaptively controls in combination with information on availability of the state return function. The control logic means preferably performs this control on a frame-by-frame basis.

  The state return means provides a state return function. The state recovery function repairs delay loss and improves pure error concealment. This function is described in US Pat. No. 6,721,327.

  If the frame is not received either because it is lost or not received in time (ie, received too late), the error concealment means will be activated and an error concealment will be attempted. On the other hand, using error concealment results in an incorrect start condition for subsequent frames. Frames that are received but not received within a time that is useful for synthesis can still be used to correct the state at the frame boundary, after which subsequent frames are decoded and synthesized. This is performed using the state of the decoder that returns to the state before the delay loss by performing additional decoding according to the state recovery method. Decode using the correct receive parameters to get the correct decoder state. Audio samples from additional decoding are discarded because they are too late to play out. The state after the error concealment frame is either replaced or combined with the state from the additional decoding to create an improved state that is more appropriate for subsequent frames. This can reduce the error propagation time.

  State recovery is used in implementations due to decoding complexity constraints, even if multiple consecutive delay losses occur, but state recovery is used in implementations, resulting in overload due to decoder complexity (and then CPU overload) It would be preferable to handle only a single or very low delay loss. FIG. 3 illustrates the benefits that state recovery provides. The upper graph of FIG. 3 discloses an undistorted waveform, where the middle waveform is distorted by delay loss and the lower waveform is distorted by delay loss, but is restored by state recovery. It should be noted that the waveform and time are slightly different because they include time measurements. The middle graph call can then be seen to decay and distort over a long period of time, resulting in poor call quality. By making the system more robust against delay loss in this way, state recovery improves performance, but increases decoding complexity by requiring additional decoding.

  The method and apparatus of the present invention gives an improvement in speech quality during the rising phase of the jitter buffer means. The rising phase of the improved jitter buffer means is described in FIGS.

  The graphs of FIGS. 4 to 6 show the level of the jitter buffer means on the vertical axis and the time on the horizontal axis. Indicates initial buffer time, rise time and full level. The initial buffer time is the time before the frame is further transmitted to the decoder (or the size of the received frame in the buffer), and the rise time is the time required to reach the full level of the jitter buffer. In FIG. 4, the dashed line indicates the fullness level of the buffered jitter buffer with improved interoperability. The solid line indicates the full level of the jitter buffer according to the method according to the invention, and the control logic means controls the initial buffer time and the time measurement that affects the rise time. This control is based on the delay loss probability during the rise time and state recovery.

  State recovery further strengthens the receiver against delay loss, so the state recovery can further reduce the initial buffer time. The control logic means according to the present invention allows adaptation of the reduced initial buffer time based on the presence / absence of state return. Since the initial buffer time can be further reduced, an improvement in interoperability is felt more than that performed in the prior art.

  As described above, the goodness of the time measurement operation depends on the acoustic characteristics of the decoded call frame. For some sounds, time measurement introduces distortion, and for certain sounds, time measurement works very well. The analysis of the acoustic properties can be used by the control logic means according to the invention to determine the aggressiveness of the time measurement, ie adapt the time measurement to the current situation. A very aggressive time measurement makes it possible to have a very short rise time of the jitter buffer means, which reduces the risk of encountering a delay spike. Active time measurement is shown in FIG. A short rise time is beneficial because state recovery is a method of concealing errors, resulting in a good but not perfect state and meaning that delay loss can still affect performance. If the channel has severe delay jitter characteristics and state recovery is not available, then it is necessary to use a very aggressive time measurement and increase the contents of the jitter buffer means very quickly. If state return is available, then the aggressiveness can be controlled according to the good time measurement execution for the current call segment. For sounds where the time measurement does not work very well, the control logic means will activate a less aggressive time measurement, thereby giving a longer rise time. Examples of various parameters or measurement scales that describe the goodness of time measurement execution and can be used by control logic means alone or in combination with other parameters / measurement scales to control the aggressiveness of time measurement are Error, energy difference between signals before and after time measurement operation, and pitch alignment error. This is shown in FIG. In this case, the control logic means preferably enables the state return function to reduce the influence of delay loss.

  Since both channel characteristics and speech signals vary over time, it would be beneficial to have control logic means that adapts between the above-described jitter buffer means start-up strategies shown in FIGS. Furthermore, having a short rise period is beneficial if the channel is changing rapidly, in which case a short rise period reduces the risk of encountering one or several delay spikes. This means that channel behavior statistics must be collected, for example from jitter buffer means, so that the statistics can be used by the control logic means to adapt the time measurement.

  Since additional decoding operations are performed, state recovery introduces extra decoding complexity that would be a heavy CPU load. The extra decoding operation is necessary because the decoder is returned to the state taken before the loss of delay and received correctly, but decoding is performed using a parameter that is delayed. The number of extra decoding operations is proportional to the delay of the delayed frame. If the frame is delayed by one frame, one extra state decoding is required. To reduce the complexity, it is not necessary to operate the synthesis filter and the subsequent filter. The synthesis filter and the subsequent filter therefore do not return. This is possible because the purpose of state recovery is only to recover a state that does not have a state return in other methods and requires a long time for repair. This includes the parts (pitch gain, pitch delay, fixed codebook gain, fixed codebook) included in the update of the adaptive codebook. This means that the increasing complexity is roughly halved.

  Extra ECU decoding is required to avoid discontinuities between the previous error concealment frame and the new good decoding frame that is decoded using the decoder's return state. To provide a smooth transition between the two decoded signals (overlap and add), an overlap period of about 5 to 20 ms is required. Thus, state recovery increases the decoding complexity and hence the CPU load. Therefore, the total complexity may exceed what the CPU can handle. It is therefore necessary to control the decoding complexity and CPU load. The control logic means according to an embodiment of the present invention retrieves information related to the CPU load and knows when the state return means should be enabled / disabled by the CPU load.

  Moreover, the use of timekeeping also introduces increased complexity and consequently increased CPU load. The control logic means monitors the total complexity used by the time measurement means, for example if the time measurement means use many of the resources that can limit the state return to a small number of parameters, or the time measurement means is subject to rough analysis. If it is found to be executed, the complexity used by the state return means is adjusted. Alternatively, the length of overlap in a synthesis that mixes operations can be reduced. The control logic means can even adjust the call parameters used in normal decoding, simplifying the synthesis phase (eg, forcing the use of integer pitch delays or even closing the ACB triggered extraction at all).

  Strict control over the use of complexity for various receiver parts requires the cycle to provide the highest possible interoperability within a given minimum call quality and a given maximum complexity margin By the way, the receiver part can use the cycle. This control is useful, for example, in embedded systems that strictly limit cycles within a cell platform. It should be noted that the complexity limit can also be done in a system, eg, a media gateway (MGW). Thus, the information related to the CPU load to be taken out can also relate to the CPU load of the MGW or the CPU load of another system.

Therefore, the present invention receives the input frame or packet, buffers it, and connects it to jitter buffer means for extracting the data frame from the received packet. It relates to time measuring means for adaptively playing out frames. Control logic means information and control logic means jitter buffer means regarding the state recovery function is available, retrieve the at least one parameter from at least one of the time measuring means and decoding means, from the jitter buffer means at least one The initial buffer time of the jitter buffer means based on information on availability of parameters and state recovery function, and the time measurement means based on information on availability of at least one parameter and status recovery function retrieved from the time measurement means or decoder Further comprising adaptively controlling at least one of the time measurements. The control logic means is preferably implemented in the receiver of the VoIP client.

The invention also relates to a method. The method includes the following steps:
1. Obtaining information regarding whether the state recovery function is available;

2. Extracting at least one parameter from at least one of the jitter buffer means, the time measurement means and the decoder, and based on the at least one parameter from the jitter buffer means and information on availability of the state recovery function, the initial buffer time of the jitter buffer means And adaptively controlling at least one of the time measurement quantities of the time measurement means based on at least one parameter retrieved from the time measurement means or decoder and information on availability of the state recovery function.

  The method can be implemented by a computer program product. Such a computer program product can be loaded directly into a computer processing means and includes software code means for performing the steps of the method.

  The computer program product can be stored on a computer-usable medium and includes a readable program that causes the computer processing means to control the execution of the steps of the method.

  In the drawings and specification, there have been disclosed typical and preferred embodiments of the invention and specific terminology has been used, but the terms are used in a generic and illustrative sense only and not for purposes of limitation. The scope of the invention is indicated by the appended claims.

It is a graph which shows operation of a time measurement function. Fig. 2 shows control logic means of a receiver according to the invention. It is a graph which shows the improvement of the performance by a state return. The function for improving the rise of the jitter buffer means according to the present invention will be described. The function for improving the rise of the jitter buffer means according to the present invention will be described. The function for improving the rise of the jitter buffer means according to the present invention will be described.

Claims (24)

Jitter buffer means for receiving and buffering input frames or packets and extracting data frames from the received packets;
Decoding means connected to the jitter buffer means for decoding the extracted data frame;
Time measuring means for adaptively playing out the decoded call frame;
Control logic means connectable to
The control logic means comprises means for obtaining information indicating whether a function of concealing packet delay is available;
The control logic means includes
Extracting at least one parameter from at least one of the jitter buffer means, the time measuring means, and the decoding means;
When the function for concealing the delay of the packet is available based on the at least one parameter from the jitter buffer means and the information indicating whether the function for concealing the delay of the packet is available Is adapted to adaptively control the initial buffering time so that the initial buffer time of the jitter buffer means is smaller ,
Based on the at least one parameter extracted from the time measuring means or the decoder and the information indicating whether a function of concealing the delay of the packet is available, the time measurement amount of the time measuring means Control at least one adaptively ,
The control logic , wherein the at least one parameter is used to determine the aggressiveness of the time measurement when adaptively controlling at least one of the time measurement amounts of the time measurement means. means.   2. The control logic means according to claim 1, wherein the parameter extracted from the jitter buffer means is related to channel characteristics.   3. The control logic means according to claim 1, wherein the parameter extracted from the decoding means is related to acoustic characteristics.   4. The parameter according to claim 1, wherein the parameter extracted from the time measuring means relates to at least one of acoustic characteristics, distortion information, and time measurement to be achieved. Control logic means. To retrieve additional parameters related to CPU load,
5. The apparatus according to claim 1, wherein at least one of an initial buffer time of the jitter buffer means and a time measurement amount of the time measurement means based on the extracted parameters is further adaptively controlled. Control logic means given in any 1 paragraph.   The information related to the load on the CPU is extracted, and the state return means is adaptively controlled based on the information related to the load on the CPU. The control logic means described.   7. The control logic unit according to claim 1, wherein the information to be extracted related to the load on the CPU is associated with a time measurement operation.   8. The control logic unit according to claim 6, wherein the information to be extracted related to the load on the CPU is associated with a time return operation.   9. The control logic unit according to claim 6, wherein the state return unit is adaptively enabled or disabled based on information related to a load on the CPU. .   9. The state return is adaptively limited to a small number of parameters, or is performed by a rough analysis based on information related to the load on the CPU. The control logic means described.   2. The adaptive control of at least one of the initial buffer time of the jitter buffer means, the time measurement means, and the time measurement amount of the state return means on a frame-by-frame basis. The control logic means according to claim 10. Jitter buffer means connected to a decoder that receives and buffers an input frame or packet, extracts a data frame from the received packet, and decodes the extracted data frame;
Time measuring means connected to the decoder for adaptively playing out decoded speech frames;
A method of controlling
-Obtaining information indicating whether a function of concealing packet delay is available;
Retrieving at least one parameter from at least one of the jitter buffer means, the time measurement means, and the decoder;
A function for concealing the delay of the packet is available based on the at least one parameter from the jitter buffer means and the information indicating whether the function for concealing the delay of the packet is available; If necessary , adaptively controlling the initial buffering time so that the initial buffer time of the jitter buffer means is smaller ;
- on the basis of said information the indicating at least one parameter, or function of hiding the delay of the packet is available to retrieve from said time measuring means or the decoder, at least for the time measurement of the time measuring means Adaptively controlling one step, and
I have a,
In the step of adaptively controlling at least one of the time measurement quantities of the time measurement means, the at least one parameter is used to determine the aggressiveness of the time measurement. .   13. The method of claim 12, wherein the parameter retrieved from the jitter buffer means is related to channel characteristics.   14. A method according to claim 12 or claim 13, wherein the parameters retrieved from the decoder are related to acoustic characteristics.   15. The parameter according to any one of claims 12 to 14, wherein the parameter extracted from the time measuring means is related to at least one of acoustic characteristics, distortion information, and time measurement to be achieved. Method. Retrieving further parameters relating to the CPU load;
-Adaptively controlling at least one of an initial buffer time of the jitter buffer means and a time measurement amount of the time measurement means based on the extracted parameter;
The method according to any one of claims 12 to 15, further comprising: Retrieving information relating to the CPU load;
-Adaptively controlling the state return means based on information relating to the load on the CPU;
The method according to claim 12, further comprising:   The method according to claim 12, wherein the information to be extracted related to the load on the CPU is associated with a time measurement operation.   19. A method according to claim 17 or claim 18, wherein the retrieved information related to the CPU load is associated with a time recovery operation.   The method according to any one of claims 17 to 19, wherein the state return means is adaptively enabled or disabled based on information related to a load on the CPU.   20. The method according to any one of claims 17 to 19, wherein the state return is limited to a small number of parameters or is performed by a rough analysis based on information related to the CPU load. . A step of adaptively controlling at least one of an initial buffer time of the jitter buffer means and a time measurement amount of the time measurement means and the state return means on a frame-by-frame basis. The method according to any one of claims 12 to 21.   23. A computer program that can be directly loaded into the internal memory of a computer in a receiver included in a packet-based communication system, wherein each step of the method according to any one of claims 12 to 22 is executed on the computer. A computer program having a software code portion to be caused.   23. Execution of each step of the method according to any one of claims 12 to 22, which is a computer program stored on a computer-usable medium and which is stored in a computer in a receiver of a packet-based communication system. A readable computer program that controls the computer.